Dr. Edgell proposes to construct and analyze libraries designed to test hypotheses concerning helix capping. A key element in the proposal is the use of regression analysis of activity data to determine factors that contribute to helix stability, including amino acid propensities for the residues in some of the helix capping motifs, and costs for motif violation. Combinatorial libraries containing randomized residues, residues constrained to be hydrophobic or hydrophilic, or to match their patterns of a target structure have been used to analyze the structure determining elements in proteins, and have been used to assess hypotheses of structure-function relationships. This project is based on two extensions on the use of combinatorial libraries. The first is the use of a regression analysis of library data, and the second is the use of split resin technology to allow the construction of arbitrarily complex libraries. The underlying assumption of this work is that regression analysis is capable of identifying and characterizing protein structural determinants. The large number of variants in a library that meet the criteria of the hypotheses to be tested will be generated using split resin technology. Whether the propensities in one instance of a motif are the same as in another, and the costs to a protein of mutations that change the motif from one "legal" type to another "legal" type will also be examined.